4 ms·

How do you crack the code to a lost ancient script?

99 points by yread 7y ago

pronoiac 7y ago
This isn’t about computer scripts, btw. > Dr Davis is talking about solving Linear A, the undeciphered language of the ancient Minoan civilisation of Crete that flourished around 1700 BCE to 1490 BCE.
meitham 7y ago
For a second I thought this was AskHN!
NotSammyHagar 7y ago
This is a really excellent article. If you ever wondered how they work out the parts of an ancient written language from basic principals, this shows you how. It's one of the best things I've ever written, I can follow the basic ideas and understand some of the things they are figuring out.
- quietbritishjim 7y ago
  Not trying to be pedantic, I'm really a bit confused here: > It's one of the best things I've ever written Do you mean one of the best things you've ever read? Or did you just call your own work "really excellent"?
  - close04 7y ago
    > and understand some of the things [off topic] I'd go with "read" :). I'd very much prefer if an author called their own work "excellent" rather than admitting they understood some of the things they wrote about, after rereading their own work. [on topic] This kind of articles makes me think how (in)adequate we are to decipher an alien transmission and language. We'll probably have to rely on "their" vastly superior intellect to make the language understandable for us. We still struggle to understand languages spoken by other humans basically identical to us, whose language evolved in conditions perfectly familiar to us (planet, environment, common concepts, etc.), and lived a mere 3500 years ago.
    - NotSammyHagar 7y ago
      Again I'm sorry, I made a mistake, I was not the author. I meant to say "it was one of the best articles I had read".
  - NotSammyHagar 7y ago
    I made a mistake, that was meant to be "I have read" about this work. I wasn't the author, sorry.
cyborgx7 7y ago
E is the most used letter, right? Just find the symbol that appears the most often and that is probably an E. /s
- HenryBemis 7y ago
  In English language, yes. But this is not even Greek, or Ancient Greek. Also I am not sure if the "icons" are letters (a, b, c, α, β, γ...) or ideograms. Good luck decyphering this without any kno I will assume that some contect is needed to start translating the text. E.g. if the tablets are found in a temple, you can start with a vocabulary of 1000 words and then expand the search. There could (?) be a "mixed" approach (technology, history, linguistics) rather than "scan and let the machine combine".
- emmelaich 7y ago
  Close; you could use Zipf's law, recently discussed here on HN. It applies pretty much to all languages. https://en.wikipedia.org/wiki/Zipf%27s_law https://en.wikipedia.org/wiki/Zipf%27s_law
jonny383 7y ago
Nice piece, great for someone with no knowledge in this domain. It's a shame even Melbourne University has been overtaken by SJW's [0]. [0] https://www.theage.com.au/national/victoria/senior-melbourne-university-academic-was-stood-down-for-bullying-20191028-p5352r.html https://www.theage.com.au/national/victoria/senior-melbourne...
romaaeterna 7y ago
There are serious problems with the Linear B decipherment from a cryptographical standpoint and a Greek standpoint. Ventris' matrix method was impossible, first of all. It didn't come close to reaching the actual search space. And the character mapping is so incredibly loose that you can translate Greek from any random collocation of linear B characters. Some of the words that Chadwick and Ventris pulled out seem to be unexpected forms for early Greek, compared to linguistic predictions. And the meaning of the texts is often fantastic compared to similar Mesopotamian tablets. My own opinion is that Ventris and Chadwick were mistaken about the decipherment, and that all of the current work on it is so much garbage. Saul Levin's 1964 "The Linear B Decipherment Controversy Re-Examined" remains the best critical examination of the matter, which should tell you something about the speed of this field. It is well past time to replicate Ventris' work with a computer approach and see if there is really any signal there.
- thaumasiotes 7y ago
  How much of a Linear B corpus is there? Deciphering Mesopotamian tablets included some experiments in which different people were given the same text to translate, and their agreement on the meaning was taken as evidence that the decipherment was more or less valid. As long as there's enough Linear B material that we don't need to assume everyone in the field will be familiar with every extant text, it should be fairly straightforward to run that same experiment for Linear B, which would pretty well settle the "the character mapping is so incredibly loose that you can translate Greek from any random collocation of linear B characters" issue. If you really want to stand behind that wording, you could very easily make up a text ("any random collocation of linear B characters") and see what different people thought it meant.
  - coldtea 7y ago
    >How much of a Linear B corpus is there? "There were 20,000 examples of Linear B signs occurring in inscriptions, compared to just 7,000 examples of Linear A signs"
  - romaaeterna 7y ago
    I don't think that you could get away with it. There's not enough out there. However you'll probably appreciate this similar jest from Michael Stokes (source Douglas Young's "Is Linear B Deciphered" for Arion in 1965). https://imgur.com/GPdnYfU https://imgur.com/GPdnYfU Article source: https://www.jstor.org/stable/20162981 https://www.jstor.org/stable/20162981

99 points by yread 7y ago

This isn’t about computer scripts, btw. > Dr Davis is talking about solving Linear A, the undeciphered language of the ancient Minoan civilisation of Crete that flourished around 1700 BCE to 1490 BCE.

meitham 7y ago

For a second I thought this was AskHN!

NotSammyHagar 7y ago

This is a really excellent article. If you ever wondered how they work out the parts of an ancient written language from basic principals, this shows you how. It's one of the best things I've ever written, I can follow the basic ideas and understand some of the things they are figuring out.

quietbritishjim 7y ago
Not trying to be pedantic, I'm really a bit confused here: > It's one of the best things I've ever written Do you mean one of the best things you've ever read? Or did you just call your own work "really excellent"?
- close04 7y ago
  > and understand some of the things [off topic] I'd go with "read" :). I'd very much prefer if an author called their own work "excellent" rather than admitting they understood some of the things they wrote about, after rereading their own work. [on topic] This kind of articles makes me think how (in)adequate we are to decipher an alien transmission and language. We'll probably have to rely on "their" vastly superior intellect to make the language understandable for us. We still struggle to understand languages spoken by other humans basically identical to us, whose language evolved in conditions perfectly familiar to us (planet, environment, common concepts, etc.), and lived a mere 3500 years ago.
  - NotSammyHagar 7y ago
    Again I'm sorry, I made a mistake, I was not the author. I meant to say "it was one of the best articles I had read".
- NotSammyHagar 7y ago
  I made a mistake, that was meant to be "I have read" about this work. I wasn't the author, sorry.

cyborgx7 7y ago

E is the most used letter, right? Just find the symbol that appears the most often and that is probably an E. /s

HenryBemis 7y ago
In English language, yes. But this is not even Greek, or Ancient Greek. Also I am not sure if the "icons" are letters (a, b, c, α, β, γ...) or ideograms. Good luck decyphering this without any kno I will assume that some contect is needed to start translating the text. E.g. if the tablets are found in a temple, you can start with a vocabulary of 1000 words and then expand the search. There could (?) be a "mixed" approach (technology, history, linguistics) rather than "scan and let the machine combine".
emmelaich 7y ago
Close; you could use Zipf's law, recently discussed here on HN. It applies pretty much to all languages. https://en.wikipedia.org/wiki/Zipf%27s_law https://en.wikipedia.org/wiki/Zipf%27s_law

jonny383 7y ago

Nice piece, great for someone with no knowledge in this domain. It's a shame even Melbourne University has been overtaken by SJW's [0]. [0] https://www.theage.com.au/national/victoria/senior-melbourne-university-academic-was-stood-down-for-bullying-20191028-p5352r.html https://www.theage.com.au/national/victoria/senior-melbourne...

romaaeterna 7y ago

There are serious problems with the Linear B decipherment from a cryptographical standpoint and a Greek standpoint. Ventris' matrix method was impossible, first of all. It didn't come close to reaching the actual search space. And the character mapping is so incredibly loose that you can translate Greek from any random collocation of linear B characters. Some of the words that Chadwick and Ventris pulled out seem to be unexpected forms for early Greek, compared to linguistic predictions. And the meaning of the texts is often fantastic compared to similar Mesopotamian tablets. My own opinion is that Ventris and Chadwick were mistaken about the decipherment, and that all of the current work on it is so much garbage. Saul Levin's 1964 "The Linear B Decipherment Controversy Re-Examined" remains the best critical examination of the matter, which should tell you something about the speed of this field. It is well past time to replicate Ventris' work with a computer approach and see if there is really any signal there.

thaumasiotes 7y ago
How much of a Linear B corpus is there? Deciphering Mesopotamian tablets included some experiments in which different people were given the same text to translate, and their agreement on the meaning was taken as evidence that the decipherment was more or less valid. As long as there's enough Linear B material that we don't need to assume everyone in the field will be familiar with every extant text, it should be fairly straightforward to run that same experiment for Linear B, which would pretty well settle the "the character mapping is so incredibly loose that you can translate Greek from any random collocation of linear B characters" issue. If you really want to stand behind that wording, you could very easily make up a text ("any random collocation of linear B characters") and see what different people thought it meant.
- coldtea 7y ago
  >How much of a Linear B corpus is there? "There were 20,000 examples of Linear B signs occurring in inscriptions, compared to just 7,000 examples of Linear A signs"
- romaaeterna 7y ago
  I don't think that you could get away with it. There's not enough out there. However you'll probably appreciate this similar jest from Michael Stokes (source Douglas Young's "Is Linear B Deciphered" for Arion in 1965). https://imgur.com/GPdnYfU https://imgur.com/GPdnYfU Article source: https://www.jstor.org/stable/20162981 https://www.jstor.org/stable/20162981